Kitchen Inventory Tracker

Kitchen Inventory Tracker: Project Story

Inspiration

Restaurants struggle to balance inventory, too much stock leads to waste, while too little leads to missed sales and unhappy customers. While working with data from Mai Shan Yun, we realized that even the smallest inefficiencies in inventory management could have a big financial impact. Our challenge for TAMU Datathon 2025 was to build a tool that would turn raw restaurant data into visual models and actionable insights, helping chefs and managers make smarter, faster decisions.

What it does

Kitchen Inventory Tracker is an interactive dashboard that combines shipment logs, ingredient usage, and menu sales trends to provide a real-time view of inventory. Our dashboard has four overarching tabs, each one with graphs and tables representing real time data provided by the company.

Overview

Displays overall trends, including sales by menu type, individual items sold, total sales, month, and category.
Shows the number of items sold per month.
Highlights estimated top menu items based on historical sales data.

Ingredient Usage

Ranks top ingredients by usage, helping managers track high-demand items.
Allows checking usage of specific ingredients to monitor inventory more closely.

Forecast

Predicts items sold and revenue, combining historical data with future projections.
Provides ingredient forecasts, including the top 5 ingredients by predicted total usage.

Shipping

Provides shipment forecasting to plan inbound inventory efficiently.
Includes a slider for custom forecast periods, showing predicted inbound quantities for selected months.

How we built it

We built the project using Python and Streamlit, which let us create a responsive, user-friendly dashboard. We started with importing shipment and usage data given by Mai Shen Yun, cleaning it with pandas, and structuring it into tables.

Challenges

Some sales, ingredient, and shipment records were incomplete or didn’t match. Predicting future sales, revenue, and ingredient needs was tricky because patterns weren’t always consistent. Running the Streamlit app locally through VS Code caused environment errors. Using ngrok to expose the local server for sharing also created connection and latency issues. To combat this, we fixed missing or mismatched records. For technical issues, we set up a clean Python environment in VS Code on one device to reduce error as my teammates and I used different platforms to code (VS Code and Google Colab) and used ngrok carefully to share the app without connection problems.

Accomplishments

Our team successfully developed a dashboard that translated raw data into visual models with interactive features. Our final product’s goal was ease of use for the user with a focus on decision-making assistance for restaurant managers. We were also able to create a predictive restocking feature which anticipates shortages before they occur, helping reduce waste and save costs. This was all accomplished through the integration of multiple data sources allowing us to display insights and conclusions that are not readily apparent.

Lessons learned

Throughout this project, we learned the importance of data cleaning, especially when working with real-world datasets. We also gained experience in building dashboards that communicate analyses visually, presenting data in detail while ensuring the method of presentation is easy to understand.

What's next for Kitchen Inventory Tracker

Looking ahead, we plan to automatically update inventory as sales happen so that restaurant functionality is maximized and ingredient shortages are noticed ahead of time. We also want to implement AI-driven demand forecasting to predict ingredient usage with higher accuracy. Eventually, the goal is to make the dashboard more interactive, allowing managers to simulate inventory decisions and see the projected impact on costs and waste.